« Faith And Works | Main | I Am Baffled »

November 17, 2008


Yeah, I think its like the whole transaction goes away. Instead of investars putting their money in a bank account, they use it to buy the security. So the 'deposit' and 'loan' both leave the bank and are out in hte market.

Yes. And to expand on yoyo. In doing so the bank makes a higher yield at lower risk and the 'depositor cum loan buyer' makes a higher yield at higher risk.

Example, suppose loans are made at prime with a 1 point initiation fee. Suppose there is a quarter point servicing fee and that deposits pay 5%. We'll say prime is 7%.

In the old days a depositor would've made 5% with no risk of principal loss (thanks to fdic). The bank would've made 7% holding the loan minus 5% paid to the depositor= 2% plus a point in the year of initiation. A loan with a 4 year term would've yielded the bank an average of 2.25% per year.

In yoyo's scenario, the depositor earns 7% as owner of the loan minus a 0.25% servicing fee = 6.75%. The depositor carries the full risk of principal loss though. The bank earns the 1% initiation fee plus 0.25% per year in servicing fees without any risk of principal loss. All the bank has to do to be ahead is initiate one more loan and it'll be earning more than it would by holding the loan to term (1.25%+1.25%>2.25%).

Another factor that favors the bank being an initiator of loans vs a holder of loans is the capital restrictions. A bank must maintain roughly 10% capital. Therefore it could earn 2.25% x 9 plus 7% x 1 = 27.25% by leveraging it's capital (ignoring operating expenses). But as a loan initiator who is not taking any lending risk, there is no effective leverage limit. It could make 1.25% an unlimited number of times.

Yes. Great summary of the mechanics and driving issues, vijay.

I have a couple of points to add. One is that I've read that a lot of those loans are to businesses tapping lines of credit, and most of them are just sitting on the money. This is part of a general rush for cash and doesn't do anything to stimulate the economy.

Also I'm sure we're all see the hilarious powerpoint that describes how subprime worked. Like vijay points out, it was basically like that for all loans (at slightly different levels of quality but still).

Oh and a third point, by having to keep the loans on the books they are incapable of generating enough profit to offset bad bets that have already been made.

"But as a loan initiator, who is not taking any lending risk, there is no effective lending limit."

Along with no leg room on airplanes, I tip my hat to the entreprenurial free market for thinking this crap up.

What's next? Spreading smallpox via infected blankets?

As a result of securitization, loan quality no longer matters. All the originator has to worry about is that the loan perform long enough - generally three to six months or so - to get it off its books. Once the loan is sold off, if the borrower defaults it's someone else's problem.

As near as I can tell, Gramm and his buddies have created a worldwide financial screw-up so monumental and so complicated that no one really knows what happened or how, much less how to fix it. I suspect there may be a sort of resonance in the system that magnifies seemingly unrelated actions to make the situation even worse when isolated fixes are tried.

The securitized loan quality problem was a problem with the Nationally Recognized Securities Rating Organizations, both in their inability to understand what they were rating and their incentive not to look to closely at the products they were being asked to rate. Much like real estate appraisers in a small town, none of them wanted a reputation as wet blankets. We either need to make the borrowers pay for the rating services or the assignments have to be done by someone independent of who is paying the bill.

Without the careless complicity of the NRSROs, the problem would not have been allowed to get as bad as it did.

It strikes me that it is far better public policy to require all originators to bear the primary risk of the loans they write and require the NRSROs to value these loans at the combined risk. B ratings for mortgages when prices had become unreasonable and loan underwriting standards had dropped would have chilled the bubble before it got out of hand.

To my eye the problem was not at the 'bank as initiator' side. It is at the 'depositor as loan buyer side' and the 'bank as servicer side'. Loans are generally made in the bank's own neighborhood where the bank is familiar with the local conditions. A loan buyer at a distance is not familiar with the local conditions and is more apt to make a blunder. It is hard for me to blame someone who sells to a willing and eager buyer. Even assuming the seller uses good ethics and all ... when it comes right down to it ... I don't know what's in the buyers head ... maybe he knows something I don't. Buyer beware is a good adage for a reason.

// We either need to make the borrowers pay for the rating services//
Generally, the borrowers want the loan more than the bank wants to make the loan. I'm not sure that this will lead to more impartiality.

// or the assignments have to be done by someone independent of who is paying the bill.//
This is done now for appraisers. The bank commissions the appraiser but the borrower pays for it. Appraisers have been assigned some culpability in the current crisis.

"To those readers who understand this better than I do: here's how I understand this: When a bank securitizes and sells a loan, it takes its profit and is free to make another loan. As long as it can sell loans and get them off its books, it can make loans endlessly, because, essentially, it can use the money it can access by selling loans to make loans, rather than being able to draw only on its own capital. (Or, if you slice things a different way: so long as it can sell loans, it can lend the same money over and over again, since when it sells the loan, it gets it back again.) Is that at all right?"

A few points.

1. In the US, for now, more or less - new rules are coming into force next year which will bring almost all existing structures on balance sheet. In Europe and many other jurisdictions securitisations were usually not off balance sheet and banks retained the first loss exposure (for mortgages, anywhere between 1% and 5%).

2. Be wary of gross lending figures. I suspect that a large proportion of the increase in, for example, home equity loans, comes from deals being unwound and/or brought back on balance sheets (remember SIVs?). So we're not talking about new lending so much as existing lending moving out of the shadow banking system into the real banking system. Furthermore, as mikkel says, drawings on a committed loan facility will increase the total today, even though the bank made its commitment for the full amount long ago.

One of the flies in the ointment is that the servicer has the ability to require the original lender to take the loan back out of the trust on default.

So, in theory at least, the bank is putting aside reserves for each loan it sells to cover the cost of the buyback. Following this logic, a bank focusing only on residential mortgages would ultimately have a balance sheet comprised of deposits on the one side and reserves for loans made and sold on the other. (Of course, they'd probably do something really smart like invest the reserves in mortgage loan securities.) The reserve obligation should, in theory, prevent the bank from using its deposits to make an infinite number of residential loans.

For a current example of failing to set aside adequate reserves, see Countrywide.

Another probable reason for the credit crunch at the level of the debt-backed securities market is that so much of that money has either been lost or is still tied up in illiquid and crappy debt backed securities issued earlier. There is less money looking for opportunities, even if there was a restoration of confidence.

If the WSJ statistics on current lending are an accurate picture, then who is perpetuating the false meme that banks are not lending, and why are they doing it?

This is part of the story. I think another part is that direct issuance of debt is way down.

The WSJ says, for example, that junk-bond issuance is 66% down from last year. So it's hard for less-than-sterling borrowers to get money.

Also, see Calculated Risk for the gap between Treasury yields and the yield on corporate bonds. The gap measures perceived risks, so you can see that the financial markets, aside from banks, are very reluctant to lend. Those borrowers who can are going to the banks. Those who can't are stuck.

"One of the flies in the ointment is that the servicer has the ability to require the original lender to take the loan back out of the trust on default."

Not so. They can require the lender to buy back loans that breach representations and warranties (eg if there was fraud in the mortgage application, or the sold loan isn't the same one as identified in the documents). Defaulted loans remain in the trust. Indeed, that's the whole point of securitisation - buyback provisions constitute implicit support and invalidate risk transfer. Countrywide's provisions were exceptional and only covered loans whose terms were modified, not ordinary delinquent or defaulted loans. You may be thinking of whole loan sales, or Fannie/Freddie mortgage backed securities, which aren't true securitisations (they're guaranteed, so the credit risk is of the sponsor, not the assets).

The comments to this entry are closed.

Blog powered by Typepad